These discrepancies among results may be ex- plained in part by differences in the trial proto- cols. In all three trials, the wholebody temperature (as measured in the rectum or esophagus) was reduced, but the strategies for cooling and the target temperatures varied: temperature was decreased to 33 to 34°C with the use of cooling blankets in TOBY and the NICHD trial and to 34 to 35°C by means of scalp cooling in the CoolCap study. The NICHD trial recorded slightly higher temperatures in the control group than did the other two trials. In TOBY, but not the other two trials, cooling was initiated during transport to the treatment center. In TOBY and the CoolCap trial, but not the NICHD trial, pa- tients were selected on the basis of the presence of abnormalities on amplitudeintegrated electroencephalography in addition to clinical criteria. Differences in local practices for withdrawal of care may also have affected outcomes. Withdrawal was slightly more common in the control group than in the cooled group in the NICHD study but was more common in the cooled group than in the control group in TOBY; these results may partially account for the greater apparent effect of hypothermia on mortality rate in the NICHD study as compared with TOBY.